0% found this document useful (0 votes)
50 views

Back To Basics - Empirical Process Control - Part 1

Process Control Engineering

Uploaded by

Chemical.Ali
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
50 views

Back To Basics - Empirical Process Control - Part 1

Process Control Engineering

Uploaded by

Chemical.Ali
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 6
Empirical Process Control: Part 1 au problams cn aft ieee eee Wiha tye of contrl do you ned? Good process control requires more than just correctly tuning the controller. Firs, you need to understand the part ofthe process to ‘Aetermine the necessary type of control. University contrel courses typically concentrate on the mathematical completes involved in modeling «contra system, but they do not spend much ise on ‘the acrual equipment used to do the controling (2), While 1Teamed about wansfer fonctions, LaPlace transforms, and other technical detain schoo, whea | _gotmy fist job, dd not even knowhow a proportional integral-derivative (PID) controller worked! In my cas, lard the bases fom an instruction manual {ora pueomatiecontller Good process contol rues more than jst correctly toning the contol Is eral tort understand the ntracsions between the pars of the paces to Gece what ype of contro is peded. This ateledesrbes how process equipment can impact process control the external factors that affect controler tuning, and the types of tuning that an be implemented, Equipment problems can affect process contral ‘Tuning system thats being used forthe frst time requires alot of experience. Typically fa control system is beng designed fora new process, the design team ‘ill spf standard PID conto loop end rely on experienced control engineers to do tein tuning. The methods deseribed inthis artist apply only ta yen tate already in operation, ening loop with contol problems i usually easy. Most a the Une, you ean spy lok al the process plas i the conta 00m ad pinpoint the ones that are chasing a setpoint or eyeing (Figo sues rated to hardware and system cnfguration, which are crcl to ood control, maybe at the roto hee ‘contol ses. This setion dserbes some ofthe problems tht can be exuted by hardware an system configurations. The examples below assume that iq Ao s being contrlle by valve, Pressure Oversized vlves. Ifa yalve it toa lange for the pres, more Mui han needed wil passthrough the valve a the contrller ope it inereaing the Nowrate prs the setpoint. As esl, he cotrller wil then lose the valve As the Gowrate decreases pas the setpoint, the contr wil start to open the valve agin. “The low wil then shoot wp once more and the controle willagain lose the valve Tis aeson crates kind of on-and-of onto that produces series of ‘bumpin the ve plat Figure 2 isa proces plot for a system that needs t least 42 f/f of steam to achieve the dested liquid temperature atthe exif «hes exchanger, When the ‘overate approaches 42 fbr the tmperatare approaches its setpoint and the onto eins to open the valve to maintain the liquid temperature. However, the valve is cveried so when tis fst 8% open, too ch steam fans into the exchanger. The temperature apy neeate and the valve cates, When the Tigo’ ei temperature begins ofall, the eee repeats 60; TITTTTT Flowrate 8 8 %Open or Flowrate, f'/he ler Output ; Ain i a Time, min 3 ven ifthe valve doesnot lose completely bt operates sees han 20% open the vale gain i, change in output change in np willbe lrg, eich wit ‘cause ange change in the process variable (PV) fre small change he controller output (OP). Fr example, opening the vale from 50% to 51% ea change of ‘only 2% af the valve setting Ineeasng the vale ftom 20% to 1% isa change of sf the valve sting This larger gun will make the controllers job mare iia, because tha o conte the ouput more precy. “To faneton propel, valves should be sie tobe 30-70% open during noel persion. This alse depends onthe range of owas nerd by the proses. ‘the desired operational flowrate is alwaysthe sme (eno chanesin production rate, the valve could be sized o operate at 70-8o% open under normal ‘ondtions, This design wil enable clacton ofthe smallest nd cheapest valve forthe aplication, “Hysteresis. When valve opening oF closing lags behind the signal andthe valve position ends up in afferent place than intended, ts ealled hysteresis. fa ‘valve tht products 10 gl/nin ffl at 5% open opened to GoM and then lowered back to 50%; itshould give fw of 10 gal/min gai, f for example, he valves suffering fom hysteresis. in nove produces 105 gli “The amount of hysteresis hat osu inthe valve called its deadband. Sometimes deadband can be lowered hy tighening the linkage between the valve stem and the actuator. But the hysteresis cannot be corrected mechanically valve postoner may be needed. Vale positioners et asa mini contro system forthe ‘valve, measuring the postion ofthe valve stem and adjusting he actuating pressure to move the vale to the desired location. A standard vale actuator (ie, ‘without a posioner) adjusts the aetuatingpresture toast value and assumes hat the valve wll rach the desired locstion, Hysteresis can lao be overcome by th integral function in PD controls, whieh continues to move the valve inthe dese direction wll the setpoint Gow is reached ‘Stein, Valves may sticks they open an le this is elle sition As the signal inereases, the valve doesnot mow unt the pressures igh enough to ‘overcome the resistance, When it does, the valve jumps tothe new postion, Ther, 35 he signal decreases, the valve stick gain until the pressures ow enough ‘thatit jumps the opposite direction. This often occurs in valves wit packed stems when the packing is compressed oo tight the stiction s severe itil produce a series of steps inthe plt of the etal fw asthe valve suddenly opens or eosesto jump tothe ne position. tn many ‘ass, ston isnot severe enough as to produce this stepwise pattern Instead, valves stick ia «postion unl dhe contr applies enough pressure thatthe ‘ave suddenly moves The valve typlelly then ale to move round well util the controller stops moving it and it tek agai (Figure) ‘ TIT 20 ie RUIUIN 1} aS za LAN YA ‘%Open or Flowrate, gal/min o. troller Output 2030 + 40 ~«50 60 70 80 90 100 ime, min rode sup paca in the poe low pls How nor ahun pron pan fasta ales shun eth ab cnecomas the essance, noes tan er tk acs In thoplotin igure, the lw isintally about 7 gal/min and fling, but the eontoller wants the lw to nereas to 79 gal/min, Starting at two minutes, the ‘controler increases the outputto the valve fom 30% to g8% and nothing happens. The, the valve stem breaks loos from the grip ofthe packing and opens to ‘38%, which suddenly increases the lw fom 75 gal/min to 90 gal/min. The contoller immediately starts eying to dose the valve to ge the Bow back below 89 ‘l/min. The wastuck valve is able to move fely in response tothe controller bringing the Low back down to around 75 gal/min, The ave stops moving and ‘becomes stuck when is about 30% open at 23 mi, The controller stat tying to open the valve anee more and te eyle repeat. “To identify ststion i problem, lok fora sudden increas inthe flowrate ae the OP has heen creasing fora wil. fan increase in OP causes the flowrate to increase slowly, the problems jus along response time nthe eontol loop, ven fhe flowrate ntl oes not respond fo several minutes. the flowrate ‘remains unchanged for several minutes and then suddenly increases, the problem isa sticking valve, valve suffering fom sition is dificul to tune and usualy requires « mechanical solution if the amount of variation It produces i problem fo the proces Lubricating the vuve stem ray inate sticton Slightly Tooening the nut chat confines the packing to reduce the presure onthe tem may also work, but the ‘valve nerds to be monitored to ensure there oleskage around the tem, ‘Wrong wale type. The thre main types of contro valves include quick-opening, linear, and equal-prcentage valves (Figure 4). Quick-opeaing valves provide large opening at low signal values; 2 higher sigoal values, the opening size increases more slow. Equal-pecentage valves do the opposite — they provide a small opening at low signal values, and asthe signal increases, the opening sz ineeases more quick. Linear valves open in proportion tothe signa value ‘throughout she range 100; 90] sof Quick ‘Opening 70) 60 50 40 30 Linear Percent Open, % Equal 20 Percentage 10] 0 0 10 20 30 40 50 60 70 80 90 100 Stem Position, % ‘Fir 4 Ths pin show held incon udp, wn-pacenags, ten cera ae ‘fa quik-opening valve i run at 25% open contro willbe difical because mal change nthe valve poston wil produce lange change nthe foe. The esl ‘uielopening valve controls best nthe ange of 40-70% apen. The det equal-perentage valve controls beet in the range of 30-60% open. These types of ‘valet are suitable for appleations where you might need many ines the normal fw in an emergency. Fr entmpl, the equa-percentage ale could be ued to ‘contol he temperature ofa ealo. The valve might normally operates 30% open, bul if he lemperaare starts orapidlyiezease the valve can be opened to 70%, proving high fate of ooking water tothe jacket to lower thereat’ temperate Linear valves povie god control across broad range of ows. They are suitable for ystems that need to operate at range of produto ates, “Interactions Between pumps and valves. There re many diferent types of pump. Some are postive displacement pumps, sich as lbe, ube and piston pumps. ‘thers are centagel pumps, which spin the gud around ina chamber to erate the pressure needed to het ow through a pipe, Cente pumps present special hind of problem for control tems because asthe Now increases, the deivng preaure tat provide the fw decrease Figure). 0 5 10 5 20 25 30 Flowrate, gal/min ‘igure 5. a ypcalhend Home cue oa carp pp Muster hg fowate he achrbl ead dopo ‘Consider a process that require flowrate of 20 gal/min and the pressure the pipe 0 psi. A valve could be sized eo achew this lowrat with a pressure rop of 15 ps (Le, half ofthe ine pressure t 50% open. This seems reasonable, uti centrifugal pump provides the pressure to drive the flow, it may not be ableto achieve the higher owrates required during upsets (eg, startups, shutdowns changes in production rte. Assume that the 3o psig pressures equal to ‘bead of about 66 ofiquld. The pump generates lower head pressures at higher flowrate. Therefore, opening the valve to increase the flowrate from 20 g/min to 25 gal/min would lowe the avalable head from 66 to 53 t— a decease of 208. the valves to operate at 90-100% open to achieve he desired Nowrat, ‘the proces wl be very diel to contol Aare, eqal-percetage valve with low pressure drop would bean appropriate valve fortis service because theincreasing rat at which this ale opens ‘would help to offset the decreasing pressure provided by the pump. Fran existing stm the contra issues could be addressed by recommending a larger valve ‘wim ora lrg impeller fo the pump. Modifying the pump wit larger impeller would flatten the pump curve at 20 gal/min an could help to stabilize ‘operation. However, this wil key increase energy oss 5 centitgal pumps typically operate most efficent in the region tothe right ofthe fit portion ofthe bead curve Look for external causes first ‘When piece of equipment has contra problems, fie examine he iputs tothe system. Control problems in column o reactor ould be duet problems upstream I, fr example the feed toa cola othe fw of tear othe ebiler aAueuatng then the contol yee cannot find a good contra pln and ‘il alway he watching fora tlation. Lok or variations inthe fed ate, ed eampaition, feed temperature sytem preare,seam presre,chilled wer twonperature, et. These external variations may he easing the problem, s they shold be adressd first One the upstream variations ae fixed, the eontol problem inthe olan reactor may resolve tlt ‘none system thet tuned found thet en entire section of the process seemed to have control problems that caused cysing. traced to acolumn thet had & ean oad and tray temperature that wer constantly osilating, checked the inputs to the columa and found thatthe feed at varied significantly. 1 traced the ‘ed ow variation back to surge tank that was collecting the bottoms from a stripping column further upstream. “Th eel inthe tank wa ekg ih conealie at 50% but there was no eat that it had to beheld at prectly that evel The level eld wary along asthe ‘anki not empty ou or overfow: loosened the level control onthe tank to allow the lve! to vary between 40% and 60%. The valve woul now only make sll aostments asthe level os rfl whic allowed sto smooth ost the variations in the tank’ discharge flowrate and limite the bi ings the ‘column's fee lowrate, Once the fed tothe column was steady the contro problems wereelimineted and the cysing sues downstream were solve, voter source of conte oben isthe setup ofthe conte stem itl Sonne equipment may bavetwo or more contol systems, Distillation columns, or ‘example, typelly havea conta on the reflux rato and on the reboler best load (29). The refs ratios adjusted to contol the temperature onan upper tay to ree an overhead parity specification, nthe rebiler i adjusted to contrl te emperaare on alow tay to meets batons specication ‘his contr! scheme sounds good in theory, but iteeates a system where the two controllers are fighting each other. Ifthe upper temperature gosto high the ‘efi ratio increases to send more ofthe low-boling component back down the column to redue the tmperstare. Tis pts more loud on the bottom ofthe ‘column and causes the temperature ofthe lower tryst decrease, The rboler conto system responds by increasing the bolup and sending mote ofthe high ‘oling component beck up the caluma, This ineeass the temperature af the upper rays and the cycle starts again, ‘To sete this ght etween the two controllers, dese which output is most important. It moe erie to kep the high-boiling component out ofthe ditiate ‘orto keep the low-boiling component ut of the bottoms? keeping the low boiler out ofthe bottoms is more important, then keep the bottoms temperature ‘controler ass and loosely tne the efx ratio controle. The botom wl holt specification ad allow the top temperature to hover around the setpoint. Any ‘variation nthe composition wil be evident inthe dstilae, When this contrl method was aplied to an actual column, ellowed good automate contro ofa system tht wa o unstable it had been run manvaly or yar, ‘What type of control do you need? ‘Most people who tune controller or usea conto program to do the tuning assume tat the cbjetve isto achieve rep esponse ona single device. But hii ot always the case. There are four base type of contr: (© Rapid reponse allows dhe proces to vaty bt tempt to bring the PV back to the setpoint within a short prio of tine {© Slow response can hep solve operational ses A pieeof equipment canbe set to have lose contro, sucha the level ofthe surge tank described nthe _revious section, which allows the equipment to absorb upstream variations. (© Hybrid response seo called cascade contro is sed when contol variable tht is manipulated to affect process value (eg, temperstre, pressure, ‘owrat) has variable source, Cascade contol ean be effective, fr example if steam is used to controls process temperature, but she Nowraeof team from the header. res due to variable header pressure. Ami oop reads the lemperstare and decides how much steam is needed I then sends this Information oa secondary lap that regulates the stem Now: The secondary loop has rapid response and the main lop has x slow response ‘© Tight control maybe needed when sl deviation in parameterslike pf or temperature ean ase produ lass or dangers conditions. The emphasis ‘ot on minimising theme ofthe respons bt on minimirng the amount of variation from the setpoint, Usually, nceasng the proportion ana integra contro parameter values ean ahiee the necessary level of contol response (Derivative contra s sometimes used for tight onto, but its amore complex procedure and isnot covered inthis artee) More advanced contro shemes canbe developed tat involve material or energy balances around a piece of equpment ora part ofthe proces. Thisis necessary in cota instances, sucha to etimat the amount of unreacted ri materia eeumulting in a reactor to avoid a runaway reaction. However, such contd schemes should not be implemented if simpler contra strategy wil work Complex ystems require tha all ofthe sensor work propery. ane temperature ot Aoverat ens dis, the whole system becomes unreal baie the material or ener balance wl be nacsrat, ‘nthe other hang fa temperature or fw sensor drifts in a simple PID system, the setpoint canbe lowered or raised in thet part ofthe system and the rest of the process can conti to run until reais caa be made, Ifa sensor malfunctions completely, the PID system canbe run oa manval ut itean be repaired. ‘Simple systems are more ragged, and with proper tuning, PED systems can contol cmos ay proces In closing ‘Tis etl dsebes some ofthe tems that an cause cotel problems, the types of control that ae avaiable, and some ations to take before starting to tune In an article to fellow, 1 wil describe the basis of the PID controller and wo diferent mthods that tn be set tune the contr for ive apts. LUterature Cited © Alford, J, and 6. Buckbee, “industrial Process Canto ystems: A New Approach a Education," Chemical Bugineering Progres, 4 (12), pp. 35-42 (ee. 2020). 4 Herzog, €.,“Deveop istlation Controls During Process Design," Chemica gineering Progress, 7 (2, p. a7 (Fe. 2023. 8 Taube, M.A. etal 19 (Mar 2020, “Achieve the Benefits of Intensification through Process Control" Chemical Bnginering Progress, 7 (9), BP.

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy